Signal Processing and Feature Extraction in Markerless Telerehabilitation

Area of Science:

• Biomedical Engineering
• Rehabilitation Science
• Digital Health

Background:

• Telerehabilitation offers remote patient support and continuity of care.
• The sit-to-stand (STS) task is vital for daily activities and functional mobility.
• Markerless motion analysis systems can facilitate remote exercise guidance.

Purpose of the Study:

• To explore telerehabilitation for remote exercise guidance using the ReMoVES system.
• To develop a quantitative parametric model for evaluating the sit-to-stand (STS) task.
• To assess the reliability and accuracy of remote performance evaluation for rehabilitation.

Main Methods:

• Utilized the ReMoVES system with Microsoft Kinect for markerless human movement monitoring.
• Employed preprocessing techniques including noise reduction and automatic segmentation for feature extraction.
• Established a reference model using healthy subjects and conducted statistical analyses with healthy subjects and patients.

Main Results:

• Identified key features for remote exercise observation through statistical analysis.
• Developed a quantitative parametric model based on automatic feature extraction of poses and body movements.
• Demonstrated the homogeneity within control group sessions for reliable parameter extraction.

Conclusions:

• The ReMoVES system provides a robust method for remote telerehabilitation.
• The quantitative parametric model enables objective evaluation and adaptation of rehabilitation plans.
• Markerless motion analysis is effective for assessing exercise execution and identifying compensatory movements remotely.